[sage-support] Re: Memory leaks with LP Solvers are back

2014-12-01 Thread Nathann Cohen
Just to say that this got fixed in #17320 http://trac.sagemath.org/ticket/17320 Nathann -- You received this message because you are subscribed to the Google Groups sage-support group. To unsubscribe from this group and stop receiving emails from it, send an email to

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-12 Thread Simon King
Hi! On 2014-11-11, Peter Mueller ypf...@googlemail.com wrote: Is it needed to install an optional package first? Which? indeed, you have to install cbc-2.8.1.p0 first (if you have sage version 6.3). It takes some time to compile. Thanks! I can confirm that there is a leak, and it seems

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-12 Thread Peter Mueller
Dear Simon, Thanks for looking into this issue. I opened a ticket, it is ticket 11917 http://trac.sagemath.org/ticket/11917. (As a naive end user of Sage, I unfortunately cannot contribute anything to resolve the problem.) Best wishes, Peter Am Mittwoch, 12. November 2014 10:57:05 UTC+1

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-12 Thread Peter Mueller
Dear Simon, Thanks for looking into this issue. I opened a ticket, it is ticket 17320 http://trac.sagemath.org/ticket/17320#ticket. (As a naive end user of Sage, I unfortunately cannot contribute anything to resolve the problem.) Best wishes, Peter Am Mittwoch, 12. November 2014 10:57:05

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Peter Mueller
A tentative solution seems to be to put a `collect()' inside the loop (and a `from gc import collect' at the beginning of the program). Still, I believe that this issue deserves attention. -- Peter M. Am Dienstag, 11. November 2014 12:20:30 UTC+1 schrieb Peter Mueller: 3 years ago a memory

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Simon King
Hi Peter, On 2014-11-11, Peter Mueller ypf...@googlemail.com wrote: A tentative solution seems to be to put a `collect()' inside the loop (and a `from gc import collect' at the beginning of the program). Still, I believe that this issue deserves attention. If gc.collect() solves the issue,

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Peter Mueller
Dear Simon, you are right, and I was wrong in claiming that collect() relieves matters. Just found out that the call of collect() tremendously slows down the program. So it would take quite long until it reaches the stage where the version without collect() needed only a few minutes to consume

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Simon King
Hi Peter, On 2014-11-11, Peter Mueller ypf...@googlemail.com wrote: Am Dienstag, 11. November 2014 12:20:30 UTC+1 schrieb Peter Mueller: while True: P = MixedIntegerLinearProgram(solver=Coin) eats several GB within a few seconds!!! The same with solver=Coin, but not with solver=glpk.

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Peter Mueller
Dear Simon, I just wanted to spend some time on debugging, but I get sage: P = MixedIntegerLinearProgram(solver=Coin) Traceback (most recent call last): ... ImportError: No module named coin_backend Is it needed to install an optional package first? Which? indeed, you have to

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Dima Pasechnik
On 2014-11-11, Peter Mueller ypf...@googlemail.com wrote: Dear Sage developers, 3 years ago a memory leak in lp solvers was fixed in ticket http://trac.sagemath.org/ticket/11917. It seems that the problems are back in Sage 6.3. The code while True: P =

[sage-support] Re: Memory leaks with LP Solvers are back

2014-11-11 Thread Peter Mueller
You mention Coin twice here. Do you mean to say Gurobi once, and Coin once? Sorry, yes. The problem arises with gurobi and cbc (=Coin), but not with glpk. -- Peter -- You received this message because you are subscribed to the Google Groups sage-support group. To unsubscribe from this